摘要
Empirically, mutual fund flows depend on past performance. It is unclear, however, whether this is rational. Using experiments, we overcome measurement problems of real data. We detect two anomalies: (1) “absolute performance effect”—investors tendency to delegate money to a fund increases with performance, even when performance is uninformative; (2) “relative performance effect”—investors tendency to delegate money decreases with other funds performance, even when their performance is attributed to luck per se. We suggest two descriptive alternatives to expected utility: (1) “subjective conditional probability”—subjective probabilities deviate from Bayesian posteriors; (2) “subjective risk aversion”—a history-dependent utility function.